Remote sensing image enhancement method and system based on visual analysis

Abstract

The invention relates to the technical field of image enhancement, in particular to a remote sensing image enhancement method and system based on visual analysis. The method comprises the following steps: firstly, acquiring a plurality of preliminary connected domains of a remote sensing image through a connected domain algorithm; then, according to the color, the texture condition, the area and the shortest distance among all edge pixel points in each preliminary connected domain, obtaining a plurality of connected domains of the remote sensing image, calculating local feature indexes of the connected domains, evaluating the merging possibility, clustering and dividing earth surface feature regions, determining the pixel point window side length and the neighborhood pixel weight, and finally, obtaining the remote sensing image. And performing enhancement processing on the remote sensing image according to the weight. According to the remote sensing image enhancement method provided by the invention, the quality of the remote sensing image is effectively enhanced, the earth surface feature recognition precision is improved, the noise is effectively removed while important details are kept, the filtering result can more accurately conform to the actual features of the image, key information is kept, and the processing efficiency and the visual effect are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image enhancement, and in particular to a remote sensing image enhancement method and system based on visual analysis. Background Art

[0002] Remote sensing images play an irreplaceable role in providing information about the earth's surface. They are widely used in scientific research, resource management, and environmental monitoring. However, due to various limitations, the spatial and temporal resolutions of remote sensing images are not high, which restricts the scope of application of remote sensing images. In order to observe the objects of interest more clearly, enhancement processing becomes a necessary step. Gaussian filtering is a classic image processing technique that is widely used in the enhancement and analysis of remote sensing images. Gaussian filtering effectively removes random noise in the image by smoothing the image using a Gaussian function, while retaining edges and important features as much as possible.

[0003] In the prior art, the traditional implementation of Gaussian filtering usually fixes the window size and calculates the weight of each pixel in the window according to the Gaussian distribution. However, this method may not be flexible and efficient when processing images with different textures and details. For example, using a larger window on the edge or in areas with rich details will produce too much smoothing effect, resulting in blurred edges and loss of details; while using a smaller window in a large area of ​​uniform smooth areas will increase the amount of calculation, resulting in unnecessary over-enhancement, which will make the filtering result inconsistent with the actual characteristics of the image. Summary of the invention

[0004] In order to solve the technical problem that the remote sensing image processing in the prior art is not flexible and efficient enough, the purpose of the present invention is to provide a remote sensing image enhancement method based on visual analysis. The technical solution adopted is as follows: In a first aspect, a remote sensing image enhancement method based on visual analysis is provided, the method comprising: Step S1: obtaining multiple preliminary connected domains of the remote sensing image through a connected domain algorithm; Step S2: obtaining multiple connected domains of the remote sensing image according to the color, texture, area, and the shortest distance between all edge pixels in each preliminary connected domain; Step S3: clustering all connected domains, and each obtained cluster represents a surface feature area; Step S4: Obtain the window side length of a pixel point according to the position of the pixel point in the surface feature area where the pixel point is located, the grayscale variance in the connected domain where the pixel point is located, and the difference in local feature indexes of the connected domains on both sides of the pixel point; Step S5: Obtain the weight of each neighboring pixel in the window of each pixel according to the grayscale and gradient of each pixel and other neighboring pixels in its window; Step S6: performing enhancement processing on the remote sensing image according to the weight.

[0005] Furthermore, the step S2 specifically includes: Obtaining a local feature index of a preliminary connected domain according to the gray value corresponding to the peak of the gray histogram of the preliminary connected domain, the variance of the gray values ​​of all pixels of the preliminary connected domain, and the mean value of the gradient of all pixels of the preliminary connected domain; According to the local feature index, area, and the shortest distance between all edge pixels of any two preliminary connected domains, the possibility of merging the two preliminary connected domains is obtained; According to the possibility of merging any two adjacent preliminary connected domains, a plurality of connected domains of the remote sensing image are obtained.

[0006] Furthermore, the gray value corresponding to the peak of the gray histogram of the preliminary connected domain, the variance of the gray values ​​of all pixels of the preliminary connected domain, and the mean of the gradients of all pixels of the preliminary connected domain are all positively correlated with the local feature index of the preliminary connected domain; The absolute value of the difference between the local feature indexes of any two preliminary connected domains and the shortest distance between all edge pixels of the two preliminary connected domains are negatively correlated with the possibility of merging the two preliminary connected domains, and the area difference between the two preliminary connected domains is positively correlated with the possibility of merging the two preliminary connected domains.

[0007] Furthermore, the step S3 specifically includes: The average value of the local feature indexes of all preliminary connected domains contained in the connected domain is taken as the local feature index of the connected domain; According to the local feature index of a connected domain and the peak value of the grayscale histogram of the connected domain, a clustering feature value of the connected domain is obtained; The difference in clustering feature values ​​between connected domains is used as the clustering distance, and all connected domains are clustered to obtain multiple clusters, each of which represents a surface feature area.

[0008] Furthermore, step S4 specifically includes: When a pixel does not belong to an edge pixel, the distance between the pixel and the edge and the grayscale variance in the connected domain where the pixel is located are negatively correlated with the window side length of the pixel; When a pixel is an edge pixel, the difference in local feature index of the connected domains on both sides of the pixel is negatively correlated with the window side length of the pixel.

[0009] Furthermore, step S4 also includes: When a pixel is an edge pixel and the connected domains on both sides of the pixel are different surface feature areas, the longer the window side length of the pixel is.

[0010] Furthermore, the step S5 specifically includes: The weight of the neighboring pixel in the window of the pixel is obtained according to the difference between the grayscale and gradient product of a pixel and the first neighboring pixel in its window, and the difference between the local feature index of the connected domain where the pixel and the neighboring pixel in its window are located; According to the weight of each neighborhood pixel of each pixel, an enhanced remote sensing image is obtained.

[0011] Furthermore, the difference between the grayscale and gradient product of a pixel point and the first neighboring pixel point in its window, and the difference in the local feature index of the connected domain where the pixel point and the neighboring pixel point in its window are located are negatively correlated with the weight of the neighboring pixel point in the window of the pixel point.

[0012] Furthermore, the step S6: For each pixel in the remote sensing image, the grayscale values ​​of all neighboring pixels in its window are weighted and summed according to the weight of each neighboring pixel of each pixel. The result is used as the new grayscale value of the central pixel after filtering. The calculated weighted sum is assigned to the central pixel to obtain an enhanced remote sensing image.

[0013] In another aspect, the present invention provides a remote sensing image enhancement system based on visual analysis, the system comprising: A preliminary connected domain acquisition module is used to acquire multiple preliminary connected domains of a remote sensing image through a connected domain algorithm; A connected domain acquisition module is used to acquire multiple connected domains of the remote sensing image according to the color, texture, area, and the shortest distance between all edge pixels in each preliminary connected domain; The surface feature region acquisition module is used to cluster all connected domains, and each acquired cluster represents a surface feature region; The window side length acquisition module is used to obtain the window side length of a pixel point according to the position of the pixel point in the surface feature area where the pixel point is located, the grayscale variance in the connected domain where the pixel point is located, and the difference in local feature indexes of the connected domains on both sides of the pixel point; A weight acquisition module is used to obtain the weight of each neighboring pixel point in the window of each pixel point according to the grayscale and gradient of each pixel point and other neighboring pixel points in its window; The enhancement processing module is used to enhance the remote sensing image according to the weight.

[0014] The present invention has the following beneficial effects: by calculating the local feature index of the connected domain, evaluating the possibility of merging, clustering and dividing the surface feature area, determining the side length of the pixel point window and the weight of the neighboring pixel, and applying Gaussian filtering and smoothing, the quality of the remote sensing image is effectively enhanced, the accuracy of surface feature recognition is improved, and the noise and smooth image are effectively removed while maintaining important details. In this way, the filtering result can more accurately meet the actual characteristics of the image, retaining key information and improving processing efficiency and visual effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 A schematic diagram of a remote sensing image provided by an embodiment of the present invention.

[0017] Figure 2 A flow chart of a remote sensing image enhancement method based on visual analysis provided by one embodiment of the present invention.

[0018] Figure 3 A block diagram of a remote sensing image enhancement system based on visual analysis provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the remote sensing image enhancement method based on visual analysis proposed by the present invention, its specific implementation method, structure, characteristics and effects are described in detail as follows in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. Unless otherwise defined, all technical and scientific terms used in this document have the same meaning as commonly understood by technicians in the technical field of the present invention.

[0020] The specific scheme of the remote sensing image enhancement method based on visual analysis provided by the present invention is described in detail below with reference to the accompanying drawings.

[0021] First, please refer to Figure 2 , which shows a flow chart of a remote sensing image enhancement method based on visual analysis provided by an embodiment of the present invention, the method comprising the following steps: Step S1: Obtain multiple preliminary connected domains of the remote sensing image through a connected domain algorithm.

[0022] Among them, step S1 specifically includes: using a drone equipped with a camera to take pictures, collecting remote sensing images, preprocessing the images, and obtaining multiple preliminary connected domains of the remote sensing images through a connected domain algorithm.

[0023] More specifically, by using drones equipped with high-precision cameras to capture remote sensing images and pre-processing the images such as denoising, ground features can be further highlighted, providing a solid foundation for subsequent analysis and application. Figure 1 In the remote sensing image shown, the river is a hydrological network composed of multi-level water systems. Its shape is diverse, often in the shape of naturally curved ribbons with different widths. Due to the reflection and absorption characteristics of water bodies to light, the river is reflected in the image as a regular, blue-green image feature, with obvious color and relatively smooth texture features in the remote sensing image; various vegetation, fallow land, etc. appear in dark green and dark brown colors, and also have relatively uniform texture features.

[0024] In remote sensing images, characteristic areas such as rivers and vegetation have different characteristics. Most surface water bodies are reservoirs formed by dams in low-lying areas, which are obviously different from the surrounding objects. Therefore, it is necessary to divide remote sensing images into multiple connected domains for further analysis.

[0025] Step S2: Acquire multiple connected domains of the remote sensing image based on the color, texture, area, and the shortest distance between all edge pixels in each preliminary connected domain.

[0026] Among them, the step S2 specifically includes: obtaining the local feature index of a preliminary connected domain according to the grayscale value corresponding to the peak of the grayscale histogram of the preliminary connected domain, the variance of the grayscale values ​​of all pixels of the preliminary connected domain, and the mean value of the gradient of all pixels of the preliminary connected domain; obtaining the possibility of merging any two preliminary connected domains according to the local feature index, area, and the shortest distance between all edge pixels of the two preliminary connected domains; and obtaining multiple connected domains of the remote sensing image according to the possibility of merging any two adjacent preliminary connected domains.

[0027] More specifically, first, the remote sensing image is operated using a connected domain algorithm to obtain multiple preliminary connected domains. Since there are differences in grayscale in different surface feature areas in the remote sensing image, there are differences in the peak values ​​of the grayscale histogram of each preliminary connected domain in the remote sensing image. For example, the grayscale value corresponding to the peak value of the grayscale histogram of the connected domain in the river area should be larger, and the grayscale value corresponding to the peak value of the grayscale histogram of the connected domain in the feature area should be smaller. Therefore, the local feature index of each preliminary connected domain can be determined based on the color and texture conditions in each preliminary connected domain. In this embodiment, the mathematical calculation formula for constructing the local feature index of each preliminary connected domain can be expressed as: ; In the formula, Indicates The gray value corresponding to the peak of the gray histogram of the preliminary connected domain; Indicates The variance of the gray values ​​of all pixels in the initial connected domain; Indicates The mean value of the gradient of all pixels in the preliminary connected domain; Indicates The local feature index of the preliminary connected domain.

[0028] In the mathematical calculation formula of the preliminary connected domain local feature index constructed above, Indicates that the The gray value range corresponding to the peak value of the gray histogram of the preliminary connected domain is limited to between 0 and 1; express The grayscale discreteness of all pixels in the initial connected domain, that is, The contrast of the initial connected domain; Indicates The grayscale changes of all pixels in the first preliminary connected domain, i.e. The clarity of the first connected domain; all three are The three are all positively correlated with the local feature index of the first preliminary connected domain. The local feature index of the preliminary connected domain.

[0029] Therefore, the gray value corresponding to the peak of the gray histogram of a preliminary connected domain, the variance of the gray values ​​of all pixels of the preliminary connected domain, and the mean of the gradients of all pixels of the preliminary connected domain are all positively correlated with the local feature index of the preliminary connected domain.

[0030] In remote sensing images, due to the complexity of surface features, a certain actually continuous feature area may be mistakenly identified as two or even more connected domains, such as at the confluence or branch of a river, or at the border of two rivers. Therefore, it is necessary to calculate the possibility of merging all two arbitrary adjacent preliminary connected domains based on the local feature index of the preliminary connected domain and the distance between adjacent connected domains. In this embodiment, the mathematical calculation formula for constructing the possibility of merging any two adjacent preliminary connected domains can be expressed as: ; In the formula, Indicates and Possibility of merging the initial connected domains; Respectively represent and The local feature index of the preliminary connected domain; Respectively represent and The area of ​​the initial connected domain; Indicates and The shortest distance between all edge pixels of a preliminary connected domain; Represents the normalization function.

[0031] In the mathematical calculation formula for the possibility of merging any two adjacent preliminary connected domains constructed above, if , it means that the local feature indexes of the two adjacent preliminary connected domains are the same. However, if the local feature indexes of the two adjacent preliminary connected domains are the same, the two places will be divided into one preliminary connected domain in the preliminary connected domain construction stage. Therefore, in this formula, ; Obviously, the adjacent and There must be a distance between all edge pixels of the initial connected domain, so ; When the difference in the areas of the two preliminary connected domains is greater, it means that the preliminary connected domain with a smaller area should be merged into the preliminary connected domain with a larger area, so Positively related to ; When the and The smaller the difference in the local feature index of the two preliminary connected domains is, the more similar the features of the two preliminary connected domains are, and the more they should be merged. Negatively related to Similarly, and The smaller the shortest distance between all edge pixels of a preliminary connected domain, the closer the two preliminary connected domains are and the more they should be merged. Negatively related to .

[0032] Therefore, the absolute value of the difference between the local feature indexes of any two preliminary connected domains and the shortest distance between all edge pixels of the two preliminary connected domains are negatively correlated with the possibility of merging the two preliminary connected domains, and the area difference between the two preliminary connected domains is positively correlated with the possibility of merging the two preliminary connected domains.

[0033] After obtaining the possibility of merging any two adjacent preliminary connected domains, the preset merging threshold is 0.6, which can be adjusted according to the specific application scenario; When it is greater than the preset merging threshold, the first and The preliminary connected domains are merged, and the adjacent preliminary connected domains that meet the conditions are merged iteratively according to the merging rules until no adjacent preliminary connected domains that meet the conditions can be found. Thus, multiple connected domains of the remote sensing image are obtained.

[0034] Each connected domain represents one of the surface feature areas in the remote sensing image, including rivers, vegetation and other objects. Therefore, it is necessary to determine the relationship between connected domains, and cluster all connected domains to classify connected domains with the same surface features. Therefore, this embodiment further sets the following steps.

[0035] Step S3: Cluster all connected domains, and each obtained cluster represents a surface feature area.

[0036] Among them, step S3 specifically includes: taking the mean of the local feature indexes of all preliminary connected domains contained in the connected domain as the local feature index of the connected domain; obtaining the clustering feature value of a connected domain according to the local feature index of the connected domain and the peak value of the grayscale histogram of the connected domain; clustering all connected domains using the difference in clustering feature values ​​between connected domains as the clustering distance to obtain multiple clusters, each cluster representing a surface feature area.

[0037] More specifically, when a connected domain is merged through multiple preliminary connected domains, the average of the local feature indexes of all preliminary connected domains is used as the local feature index of the connected domain. Moreover, for different surface feature areas, the river area is usually smaller than the ground feature area, so the number of pixel points of the peak in the grayscale histogram is also different. Therefore, in this embodiment, the mathematical calculation formula for constructing the clustering feature value of each connected domain of the remote sensing image can be expressed as: ; In the formula, Indicates The clustering eigenvalues ​​of connected domains; Indicates The local characteristic index of the connected domain; Indicates The peak value of the grayscale histogram of the connected domain.

[0038] In the mathematical calculation formula of the clustering characteristic value of each connected domain constructed above, when the local characteristic index between connected domains and the peak value of their grayscale histogram are closer, they are more likely to belong to the same surface characteristic area, so when clustering, they should have a smaller clustering distance, that is, the smaller the difference in clustering characteristic values. The clustering eigenvalues ​​of connected components , No. The local characteristic index of the connected domain They are positively correlated with the clustering eigenvalues ​​of the connected domain.

[0039] Furthermore, in remote sensing images, the difference in clustering feature values ​​between connected domains is used as the clustering distance, and all connected domains are clustered to obtain several clusters. The connected domains of all different clusters are marked with different codes, for example, they can be coded 1, 2, 3..., and the connected domains in the same cluster have the same code. Based on this, all connected domains are divided into multiple surface feature areas, and each surface feature area has a code.

[0040] In order to effectively enhance the remote sensing image, it is necessary to further determine the dynamic window size for each pixel in the remote sensing image. Therefore, this embodiment further arranges the following steps.

[0041] Step S4: Obtain the window side length of a pixel point according to the position of the pixel point in the surface feature area where the pixel point is located, the grayscale variance in the connected domain where the pixel point is located, and the difference in local feature indexes of the connected domains on both sides of the pixel point.

[0042] Among them, step S4 specifically includes: when a pixel point does not belong to an edge pixel point, the distance of the pixel point from the edge and the grayscale variance in the connected domain where the pixel point is located are negatively correlated with the window side length of the pixel point; when a pixel point belongs to an edge pixel point, the difference in local feature index of the connected domains on both sides of the pixel point is negatively correlated with the window side length of the pixel point, and when a pixel point belongs to an edge pixel point, the connected domains on both sides of the pixel point are different surface feature areas, the larger the window side length of the pixel point.

[0043] More specifically, when performing Gaussian filtering on remote sensing images, when a pixel point belongs to a large smooth area within a surface feature area, a larger window should be assigned to avoid over-enhancing invalid information. When a pixel point is near an edge pixel point, that is, it belongs to edge detail information that should be greatly enhanced, a smaller window should be assigned, because edge information is crucial for the recognition and segmentation of surface features, and using a small window can better preserve and enhance these edge details. That is, the farther the pixel point in the connected domain is from the edge, the larger the window of the pixel point; and when a pixel point is an edge pixel point and the connected domains on both sides are different surface feature areas, that is, the codes of the connected domains on both sides are different, then it is an effective detail that needs to be enhanced and has a relatively small window.

[0044] In this embodiment, the mathematical calculation formula for constructing the window side length of each pixel point of the remote sensing image is as follows: ; In the formula, Represents the window side length of pixel m in the remote sensing image; Indicates the artificially preset standard window side length; Represents the minimum value of the distance between pixel m and all edge pixels in its connected domain. Obviously, ; Represents the grayscale variance of the connected domain where the pixel m is located. Since the grayscale in a connected domain is not exactly the same, ; Represents the absolute value of the difference between the local feature indexes of the connected domains on both sides of the pixel m; is a coefficient, which takes the value 1 when the codes of the connected domains on both sides of the pixel point m are the same, and takes the value 2 when the codes are different; Represents a set of edge pixels; represents the normalization function; Represents an exponential function with the natural constant e as base.

[0045] In the mathematical calculation formula for the window side length of each pixel in the remote sensing image constructed above, when the pixel m is not an edge pixel, the farther it is from the edge, the more it is in a relatively smooth area in the connected domain, and the smaller the grayscale variance in the connected domain where the pixel m is located, the more uniform the connected domain is, the larger the window is assigned and the weaker the filtering enhancement effect is; when the pixel m is an edge pixel, the greater the difference in local feature index of the connected domains on both sides of the edge pixel, the smaller the window is, and when the connected domains on both sides are different surface feature areas, the degree of enhancement should be stronger, that is, the smaller the window is.

[0046] After obtaining the window side length of the pixel point in the remote sensing image, the square of the side length is the window size of the pixel point. Further, it is necessary to accurately obtain the weights of all the neighborhood pixels in the window so as to enhance the remote sensing image by the weights of all the neighborhood pixels in the window. Therefore, the present embodiment further arranges the following steps.

[0047] Step S5: Obtain the weight of each neighboring pixel in the window of each pixel according to the grayscale and gradient of each pixel and other neighboring pixels in its window.

[0048] Step S5 specifically includes: obtaining the weight of the neighboring pixel point in the window of the pixel point according to the difference between the grayscale and gradient product of a pixel point and the first neighboring pixel point in its window, and the difference between the local feature index of the connected domain where the pixel point and the neighboring pixel point in its window are located; and obtaining an enhanced remote sensing image according to the weight of each neighboring pixel point of each pixel point.

[0049] More specifically, in general, if the features of a neighborhood pixel (such as color, brightness, texture, etc.) are highly consistent with the features of the central pixel, then the neighborhood pixel will be given a higher weight when performing image enhancement because it is more likely to contain valuable information for the central pixel. This mechanism ensures the local consistency and smoothness of image processing, so that noise can be effectively removed or other enhancement processing can be performed while retaining image edges and details. In the window of each pixel point, the more similar the features of the neighborhood pixel point are to the central pixel point, the greater its weight, and the weight of each neighborhood pixel point in the window of each pixel point is determined accordingly. In this embodiment, the mathematical calculation formula for the weight of each neighborhood pixel point in the window of each pixel point is constructed as follows: ; In the formula, Represents the pixel m in the remote sensing image within the window The weight of the neighboring pixels, , Represents the number of all neighboring pixels in the window of pixel m; Represents pixel m and its window The difference between the grayscale and gradient product of the neighboring pixels is obvious. ; Represents pixel m and its window The difference in local feature index of the connected domain where the neighboring pixels are located.

[0050] In the remote sensing image constructed above, within the window of pixel point m, In the mathematical calculation formula for the weight of neighboring pixels, pixel m and its The difference between the grayscale and the gradient product of the neighboring pixels The smaller the pixel m is, the smaller the pixel m is in the window. The difference of local feature index of connected domain where neighboring pixels are located The smaller it is, the more consistent the features of its neighborhood pixels are with the features of the central pixel. Because it is more likely to contain valuable information for the central pixel, it needs to be given a higher weight.

[0051] Therefore, the difference between the grayscale and gradient product of a pixel and the first neighboring pixel in its window, and the difference in the local feature index of the connected domain where the pixel and the neighboring pixel in its window are located are negatively correlated with the weight of the neighboring pixel in the window of the pixel.

[0052] Step S6: performing enhancement processing on the remote sensing image according to the weight.

[0053] Specifically, for each pixel in the remote sensing image, the grayscale values ​​of all neighboring pixels in its window are weighted and summed according to the weight of each neighboring pixel of each pixel, and the result is used as the new grayscale value of the central pixel after filtering. The calculated weighted sum is assigned to the central pixel to obtain an enhanced remote sensing image.

[0054] Second, this embodiment provides a remote sensing image enhancement system based on visual analysis, see Figure 3 , which shows a block diagram of a remote sensing image enhancement system based on visual analysis provided by an embodiment of the present invention, the system comprising: The preliminary connected domain acquisition module 101 is used to acquire multiple preliminary connected domains of the remote sensing image through a connected domain algorithm; A connected domain acquisition module 102 is used to acquire multiple connected domains of the remote sensing image according to the color, texture, area, and the shortest distance between all edge pixels in each preliminary connected domain; The surface feature region acquisition module 103 is used to cluster all connected domains, and each acquired cluster represents a surface feature region; The window side length acquisition module 104 is used to acquire the window side length of a pixel point according to the position of the pixel point in the surface feature region where the pixel point is located, the grayscale variance in the connected domain where the pixel point is located, and the difference in local feature indexes of the connected domains on both sides of the pixel point; A weight acquisition module 105 is used to acquire the weight of each neighboring pixel in the window of each pixel according to the grayscale and gradient of each pixel and other neighboring pixels in its window; The enhancement processing module 106 is used to enhance the remote sensing image according to the weight of each neighborhood pixel point in the window of each pixel point.

[0055] The remote sensing image enhancement method and system based on visual analysis provided in this embodiment first obtains multiple preliminary connected domains of the remote sensing image through a connected domain algorithm; then, according to the color, texture condition, area, and the shortest distance between all edge pixels in each preliminary connected domain, obtains multiple connected domains of the remote sensing image, further clusters all connected domains, and each obtained cluster represents a surface feature area, and obtains the window side length of a pixel point according to the position of the pixel point in the surface feature area where it is located, the grayscale variance in the connected domain where the pixel point is located, and the difference in local feature index of the connected domains on both sides of the pixel point. Finally, according to the grayscale and gradient of each pixel point and other neighboring pixel points in its window, the weight of each neighboring pixel point in the window of each pixel point is obtained, and the remote sensing image is enhanced according to the weight. In this remote sensing image enhancement processing method, the filtering result can more accurately meet the actual characteristics of the image, which not only retains key information, but also improves processing efficiency and visual effects.

[0056] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0057] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A remote sensing image enhancement method based on visual analysis, characterized in that: The method comprises: Step S1: obtaining multiple preliminary connected domains of the remote sensing image through a connected domain algorithm; Step S2: obtaining multiple connected domains of the remote sensing image according to the color, texture, area, and the shortest distance between all edge pixels in each preliminary connected domain; Step S3: clustering all connected domains, and each obtained cluster represents a surface feature area; Step S4: Obtain the window side length of a pixel point according to the position of the pixel point in the surface feature area where the pixel point is located, the grayscale variance in the connected domain where the pixel point is located, and the difference in local feature indexes of the connected domains on both sides of the pixel point; Step S5: Obtain the weight of each neighboring pixel in the window of each pixel according to the grayscale and gradient of each pixel and other neighboring pixels in its window; Step S6: performing enhancement processing on the remote sensing image according to the weight.

2. The remote sensing image enhancement method based on visual analysis according to claim 1, characterized in that: The step S2 specifically includes: Obtaining a local feature index of a preliminary connected domain according to the gray value corresponding to the peak of the gray histogram of the preliminary connected domain, the variance of the gray values ​​of all pixels of the preliminary connected domain, and the mean value of the gradient of all pixels of the preliminary connected domain; According to the local feature index, area, and the shortest distance between all edge pixels of any two preliminary connected domains, the possibility of merging the two preliminary connected domains is obtained; According to the possibility of merging any two adjacent preliminary connected domains, a plurality of connected domains of the remote sensing image are obtained.

3. The remote sensing image enhancement method based on visual analysis according to claim 2, characterized in that: The step S2 specifically includes: The gray value corresponding to the peak of the gray histogram of the preliminary connected domain, the variance of the gray values ​​of all pixels of the preliminary connected domain, and the mean of the gradients of all pixels of the preliminary connected domain are all positively correlated with the local feature index of the preliminary connected domain; The absolute value of the difference between the local feature indexes of any two preliminary connected domains and the shortest distance between all edge pixels of the two preliminary connected domains are negatively correlated with the possibility of merging the two preliminary connected domains, and the area difference between the two preliminary connected domains is positively correlated with the possibility of merging the two preliminary connected domains.

4. The remote sensing image enhancement method based on visual analysis according to claim 1, characterized in that: The step S3 specifically includes: The average value of the local feature indexes of all preliminary connected domains contained in the connected domain is taken as the local feature index of the connected domain; According to the local feature index of a connected domain and the peak value of the grayscale histogram of the connected domain, a clustering feature value of the connected domain is obtained; The difference in clustering feature values ​​between connected domains is used as the clustering distance, and all connected domains are clustered to obtain multiple clusters, each of which represents a surface feature area.

5. The remote sensing image enhancement method based on visual analysis according to claim 1, characterized in that: Step S4 specifically includes: When a pixel does not belong to an edge pixel, the distance between the pixel and the edge and the grayscale variance in the connected domain where the pixel is located are negatively correlated with the window side length of the pixel; When a pixel is an edge pixel, the difference in local feature index of the connected domains on both sides of the pixel is negatively correlated with the window side length of the pixel.

6. The remote sensing image enhancement method based on visual analysis according to claim 5, characterized in that: Step S4 also includes: When a pixel is an edge pixel and the connected domains on both sides of the pixel are different surface feature areas, the longer the window side length of the pixel is.

7. The remote sensing image enhancement method based on visual analysis according to claim 1, characterized in that: The step S5 specifically includes: The weight of the neighboring pixel in the window of the pixel is obtained according to the difference between the grayscale and gradient product of a pixel and the first neighboring pixel in its window, and the difference between the local feature index of the connected domain where the pixel and the neighboring pixel in its window are located; According to the weight of each neighborhood pixel of each pixel, an enhanced remote sensing image is obtained.

8. The remote sensing image enhancement method based on visual analysis according to claim 7, characterized in that: The difference between the grayscale and gradient product of a pixel and the first neighboring pixel in its window, and the difference in the local feature index of the connected domain where the pixel and the neighboring pixel in its window are located are negatively correlated with the weight of the neighboring pixel in the window of the pixel.

9. The remote sensing image enhancement method based on visual analysis according to claim 7, characterized in that: Step S6: For each pixel in the remote sensing image, the grayscale values ​​of all neighboring pixels in its window are weighted and summed according to the weight of each neighboring pixel of each pixel. The result is used as the new grayscale value of the central pixel after filtering. The calculated weighted sum is assigned to the central pixel to obtain an enhanced remote sensing image.

10. A remote sensing image enhancement system based on visual analysis, characterized in that: The system comprises: A preliminary connected domain acquisition module is used to acquire multiple preliminary connected domains of a remote sensing image through a connected domain algorithm; A connected domain acquisition module is used to acquire multiple connected domains of the remote sensing image according to the color, texture, area, and the shortest distance between all edge pixels in each preliminary connected domain; The surface feature region acquisition module is used to cluster all connected domains, and each acquired cluster represents a surface feature region; The window side length acquisition module is used to obtain the window side length of a pixel point according to the position of the pixel point in the surface feature area where the pixel point is located, the grayscale variance in the connected domain where the pixel point is located, and the difference in local feature indexes of the connected domains on both sides of the pixel point; A weight acquisition module is used to obtain the weight of each neighboring pixel point in the window of each pixel point according to the grayscale and gradient of each pixel point and other neighboring pixel points in its window; The enhancement processing module is used to enhance the remote sensing image according to the weight.

Images